treePL: divergence time estimation using penalized likelihood for large phylogenies.

Area of Science:

• Computational Biology
• Phylogenetics
• Evolutionary Biology

Background:

• The rapid growth of genetic data and advancements in phylogenetic reconstruction algorithms have led to the construction of increasingly large phylogenies.
• Existing divergence time estimation methods are often computationally limited, restricting their application to smaller datasets compared to recently published large phylogenies.

Purpose of the Study:

• To present a novel algorithm and implementation for estimating divergence times in large phylogenies.
• To address the computational challenges associated with analyzing datasets containing thousands of taxa.

Main Methods:

• Development of a divergence time estimation method utilizing penalized likelihood.
• Implementation of a hybrid optimization strategy combining standard derivative-based optimization with stochastic simulated annealing.
• Comparison of the new method's performance against established software such as r8s, PATHd8, and BEAST.

Main Results:

• The developed algorithm successfully handles large phylogenetic datasets with thousands of taxa.
• The combined optimization approach effectively overcomes computational challenges in divergence time estimation.
• Performance benchmarks indicate the method's capability in analyzing large-scale phylogenetic data.

Conclusions:

• The new penalized likelihood method provides a scalable solution for divergence time estimation in the era of big phylogenetic data.
• This approach expands the scope of phylogenetic analyses that can be performed on large, complex datasets.
• The software implementation (treePL) is available for broader scientific use.